> Not to mention, even if your supercomputer has an idea you still need a lab that can confirm the idea. I think this is the critical aspect. You don't need to

> Not to mention, even if your supercomputer has an idea you still need a lab that can confirm the idea. I think this is the critical aspect. You don't need to own a supercomputer to run simulations in supercomputers. Sometimes you don't even need supercomputers at all. It might be a challenge but it's not prohibitive. What's clearly a blocker is the verification & validation work to assess if a specific model actually works. You need to run experimental tests to check if your module is any good, and you need to run experimental tests to check if the output of your computational models correlate with reality. That costs both time and money.